Reading resolution charts comparing MFT lenses to FX lenses

Started Mar 29, 2013 | Questions thread
Dr_Jon Veteran Member • Posts: 4,788
Re: Reading resolution charts comparing MFT lenses to FX lenses

Thanks for that considered reply. It's tricky as the smaller sensors have advantages with over-sized lenses (using the "best bit" of the lens and the corners would have less of an angle for the incoming light if using the same flange distance). As to just how a lens scales from FF to crop it seems complicated as you'd really like the smaller sensor to have the same pixel count and, relatively, the same strength AA filter.

I think perhaps we misunderstood each other (sorry for my bit in that if so). I'm okay with lp/mm if not compared directly but the sensor/pixel sizes and other factors are allowed for. Perhaps my somewhat ongoing disagreement with Anders over the absolute quality of m43 vs. FF coloured my reading of your comments (plus I was a bit exhausted yesterday).

Anyway, my original reading of the OP's question was whether FF lenses on FF cameras will produce sharper pictures than m43 lenses on m43 cameras and I think they will. I think that's partly down to the rest of the image processing system and partly that Canon and Nikon do make some very sharp glass. Other opinions are available... lots of them... Whether sharp FF lenses will be better on m43 cameras than sharp m43 lenses is a more complicated one.

Comparing MTF curves is also tricky, as manufacturers don't seem keen on producing them in comparable ways. Looking at the curves for the Olympus 75/1.8, perhaps the sharpest m43 lens, at then comparing with the super-sharp Canon 300/2.8 II, (look at the lowest one, the others are with extenders), would suggest the Canon is miles better (I suspect it is somewhat better, my 300/2.8 IS is stunning and the mkII curves are a step up from it) but I'm not sure it's useful as the data isn't produced the same way.

Thinking about using lp/mm for comparison is complicated as FF lenses are usually compared using much larger pixel size cameras, limiting the lp/mm achieveable. It's perhaps not so hard to work out the FF/FF vs. m43/m43 relative sharpnesses, but tricky to work out how much better an FF lens would be on a higher resolution/mm sensor (which m43 cameras must be as long as they have higher pixel densities).

My 5DmkII has 3755 vertical pixels and the GH3 3456 pixels, that's only 8% less which isn't much. If I go to LensTip and use their lp/mm figures my 35-100 lens nearly gets to 80lp/mm at the centre at 100mm and f5.6, it's a lot less everywhere else. My Canon equivalent is above 40 lp/mm at all focal lengths and all over the frame from f2.8-f8.0. You'd think then the Canon system was sharper as it seems to be making better use of the Camera's resolution.

However unfortunately they are using a 12MP m43 sensor (ph=3024, plus Olympus AA filter/processing, 224 pixels/mm) vs a 20MP (almost exactly identical to my 5DmkII) sensor for Canon so they aren't that comparable. You then need to estimate how much of a limiting effect the camera (etc.) has on the result. Fun... err, not.

The super-sharp 75/1.8 hits 82 lp/mm on the same camera but is mostly some way below that, so perhaps the camera isn't the absolute limiting factor (if so I'd expect a very sharp lens to have more of a slightly rounded plateau). Perhaps the Camera the use is more just taking the edge off the results.

Looking at their results for the Canon lens (300mm f2.8 II) that peaks at about 47 lp/mm, suggesting that might be the limit of what the total test system can resolve in their tests (the 70-200 gets to a bit of a plateau at 48-ish).

However what it doesn't say is how much the Canon lens has left in hand, as the testing camera's pixels are about half the height of the GH3 ones so it can't say if the lens could resolve further unless someone adapts it to a GH3 and measures it. Also doesn't help with how much the GH3 would improve things over the 12MP camera. Hence hard to say what the Canon would do on a m43 camera.

Roger's tests ( show, as you would expect, that more camera/sensor resolution = better overall resolution, but nothing like linearly, again as you'd expect. I'm better at bandwidth maths on analogue systems (3dB points, etc.), but cameras throw in Nyquist and other annoyances and who knows how the test software interpolates the results.

On the above basis I think (personally) it is reasonable to say (using the above lenses) a pic on my Canon will appear sharper (at the same visual size) than a pic from my GH3 as there is more sharpness over the picture height (and believing Imatest on the merits of MTF50 as a measure of visual sharpness). That's using the 75mm result to suggest the GH3 won't add more than say 3-4% to the numbers (14% extra vertical pixels, extra resolution probably less so with educated guessing on AA filter strengths, but if the lens is the more limiting factor at 12MP then we will see well under half of that increase).

One thing I did do a while back was compare lens resolution on my 5DmkII vs a 7D to see how economical it was to use a 7D to give extra reach with a lens rather than buying the next big white lens up. A 7D has 1.5x the linear pixel density of the 5DmkII but IIRC the improvement in sharpness was more 20--30% than 50%, suggesting it isn't the way to go.

Also I'm not going too far with the analysis as the extra per-pixel noise from the m43 sensor as  would also limit the resolution as the pixel density went up, the larger sensor in vaguely comparable technology will always be ahead there.

Wrote that a bit quickly as have to go out, hope there is some sense in it...

 Dr_Jon's gear list:Dr_Jon's gear list
Fujifilm FinePix Real 3D W3 Sony RX100 V Canon EOS 5DS R Panasonic GH5 Canon EF 14mm f/2.8L II USM +26 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow